ClawHub
Back to skill

Security audit

Tesseract OCR文字识别

Security checks across malware telemetry and agentic risk

Overview

This is a straightforward OCR helper that reads a user-specified image and optionally saves OCR results, with no evidence of hidden persistence, network exfiltration, or destructive behavior.

Install only if you are comfortable running a local Python OCR helper with pytesseract, Pillow, and Tesseract installed. Use it on images you intend to process, avoid pointing it at sensitive files, and choose --output paths carefully to avoid overwriting important files.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Lp3

Medium
Category
MCP Least Privilege
Confidence
80% confidence
Finding
The skill metadata declares no permissions, but the implementation apparently has file-write capability (consistent with the documented `--output` parameter that writes OCR results to a file). This is a real security issue because undocumented write behavior weakens the trust boundary for users and policy engines: a caller may assume the skill is read-only while it can modify the filesystem, enabling unintended overwrites or writes to sensitive paths if inputs are not constrained.

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The skill accepts an arbitrary local `image_path` and opens it directly, enabling callers to read from any filesystem path the agent process can access. Even though the code is intended for OCR, this broad file access can be abused to probe local files, trigger parsing of unintended files, and exfiltrate text from sensitive images or documents outside the skill's expected scope.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.